120 likes | 319 Views
Kord Smith. Ben Forget. Ju Li. Felix Para. Sid Yip. 22.107 Computational Nuclear Science and Engineering (New) 3-0-9 H-LEVEL Grad Credit (Spring) Prerequisites: 12.010, 18.085, or permission of instructor. by Doing Nuclear Science and Engineering Problems!.
E N D
Kord Smith Ben Forget Ju Li Felix Para Sid Yip 22.107 Computational Nuclear Science and Engineering (New) 3-0-9 H-LEVEL Grad Credit (Spring) Prerequisites: 12.010, 18.085, or permission of instructor by Doing Nuclear Science and Engineering Problems! Practical programming / debugging skills Mathematical insights into how algorithms work Model construction & interpretation of numerical results
MIT NSE Undergraduate Curriculum all MIT NSE undergraduates reach basiclevelof Practical programming / debugging skills Mathematical insights into how algorithms work
Course 6: Electrical Engineering and Computer Science OR Course 12: Earth, Atmospheric, and Planetary Sciences
Course 18: Mathematics Excerpts from http://math.mit.edu/classes/18.085/ Homework 3 is due FRIDAY Sept 30:MATLAB Question : This uses the codes just added to the website for advection-diffusion -Du_xx + vu_x = 1 in the earlier hwk problem 1.2.19. Those codes use finite differences (u_x is centered or upwind) and compare with the true u(x). Part 1: Derive the true solution given in the code. A particular solution is u=x/v. Add the general solution to -D u_xx + v u_x = 0 with constants A and B, then find A and B. Part 2: Decide the accuracy (both centered and upwind) using the new code Ch1Q19FiniteDiffsConvergence.m that is also on the website The max error decreases with what power of h as h goes to 0 ? Part 3: Continue the experiment for very small h. We think the error may start to increase and we don't know why. Why ? Notes on Homework 2 - [L,U] = lu(A) includes any row exchanges in L (so it may not be triangular) [L,U,P] = lu(A) will separate out that permutation matrix - The product of pivots in U is MINUS the determinant of A after an odd number of row exchanges - derivative of delta function : Integrate by parts to see that it picks out -g'(0)- to avoid row exchanges for positive definite matrices K, use chol(K) where chol = Cholesky
But nothing systematically computational exist at the Graduate level 3 graduate core subjects: Applied Nuclear Physics (22.101) Electromagnetic Interactions (22.105) Neutron Interactions and Applications (22.106) in dissonance with the facts that • Computation has become the third method of inquiry in nuclear science and engineering, in addition to experiments and analytical theory • - Knowing how to identify and use computers to solve problems (model construction, programming, compiling, debugging, profiling, interpreting) is a key survival skill in workplace.
22.106 Neutron Interactions and Applications Prereq: 22.101 Units: 3-0-9 Comprehensive treatment of neutron interactions in condensed matter at energies from thermal to MeV, focusing on particle distributions most relevant to fission, fusion and radiation research applications. Neutron distributions in reactor, accelerator and material structures resulting from single and multiple reactions, and in wave phenomena (optics) and inelastic scattering experiments. Comparison of neutron and fluid transport. Particle simulations (Monte Carlo simulations). Term paper and presentation required. B. Forget 22.212 Nuclear Reactor Analysis II Prereq: 22.106, 22.211, permission of instructor Units: 3-2-7 Addresses advanced topics in nuclear reactor physics with an additional focus towards computational methods and algorithms (towards transport). Covers current computational methods employed in lattice physics calculations such as resonance models, critical spectrum adjustments, advanced homogenization techniques and fine mesh transport theory models. Deterministic transport approximation techniques such as the method of characteristics, discrete ordinates methods, response matrix methods and finite elements methods presented as well as adaptivity methods. Acceleration techniques for these various solution schemes and extension to 3-D core calculations discussed. Non-linear algorithms for eigenvalue problems and multiphysics coupling also covered. Requires a strong computational background and knowledge of C/C++ or Fortran. B. Forget
22.251 Systems Analysis of the Nuclear Fuel Cycle Prereq: 22.05 Units: 3-2-7 Lecture: MW1-2.30 (24-115) Study of the relationship between the technical and policy elements of the nuclear fuel cycle. Topics include uranium supply, enrichment, fuel fabrication, in-core reactivity and fuel management of uranium and other fuel types, used fuel reprocessing and waste disposal. Principles of fuel cycle economics and the applied reactor physics of both contemporary and proposed thermal and fast reactors are presented. Nonproliferation aspects, disposal of excess weapons plutonium, and transmutation of long lived radioisotopes in spent fuel are examined. Several state-of-the-art computer programs relevant to reactor core physics and heat transfer are provided for student use in problem sets and term papers. Kord Smith 22.617 Plasma Turbulence and Transport Prereq: 22.616 or permission of instructor Units: 3-0-9 Introduces plasma turbulence and turbulent transport, with a focus on fusion plasmas. Covers theory of mechanisms for turbulence in confined plasmas, fluid and kinetic equations, and linear and nonlinear gyrokinetic equations; transport due to stochastic magnetic fields, magnetohydrodynamic (MHD) turbulence, and drift wave turbulence; and suppression of turbulence, structure formation, intermittency, and stability thresholds. Emphasis on comparing experiment and theory. Discusses experimental techniques, simulations of plasma turbulence, and predictive turbulence-transport models. A. White
22.107 Course Goal by Doing Nuclear Science and Engineering Problems! Mathematical insights into how algorithms work Practical programming / debugging skills Model construction & interpretation of numerical results Intermediate Level Reached Intermediate Level Reached Intermediate Level Reached • incoming MIT NSE graduate students have diverse backgrounds • but, NSE don’t need or have time to reinvent the wheels • 22.107 Course Pre-requisite: 12.010, 18.085 • Assumes basic level of numerical linear algebra, finite difference, FFT etc. (if not confident, take 18.085) • Assumes basic level of programming skills (if not confident, take 12.010)
22.107 Course Approach • Not gonna babysit and spoon feed: lectures provide pointers (references, websites) and inspirational examples • Self study and self-motivated programming a must • Problem set centric: develop critical analysis and synthetic problem-solving skills by asking them to solve problems with fewer and fewer constraints, end course with completely open-ended term project • Arbitrary programming language: ask to show excerpts of source code and intermediate data • Have fun programming and solving problems.
Develops practical scientific computing skills with applications in radiation physics, reactor engineering and design, nuclear materials, fusion, etc. Compiling/profiling/time and memory complexities/debugging. Solvers of ordinary differential equations and partial differential equations. Error versus stability. Pre-and post-processing. Survey of visualization and parallel computing. Case studies in quantum mechanics, neutron diffusion and transport, simple CFD, and radiation cascade simulations. Homework requires programming in one or several languages of choice: some Matlab-free homeworks enforced.
Beyond 22.107… A separate Monte Carlo course (maybe 22.108) in the works, by Kord Smith Ju Li will develop the initial 22.107 offering in an “open-source” fashion… 22.107: Mainly deterministic, continuum field methods 22.108: Mainly stochastic, discrete-agents based methods